The EMBO Journal vol. 1 1 no. 10 pp.3645 - 3652, 1992

Trans regulation in the Ultrabithorax of Drosophila: alterations in the promoter enhance transvection

Antonio Martinez-Laborda, 1991), yellow (Geyer et al., 1990) and some other , Acaimo Gonzalez-Reyes and Gines Morata but it is in Ubx where it has been the most extensively studied (Lewis, 1954, 1955, 1982, 1985; Babu and Bhat, 1981; Babu Centro de Biologia Molecular (CSIC-UAM), Universidad Aut6noma et al., 1987; Micol and Garcfa-Bellido, 1988; Castelli-Gair de Madrid, 28049 Madrid, Spain et al., 1990; Mathog, 1990; Micol et al., 1990). This gene Communicated by G.Morata offers several advantages for the study of transvection. (i) the genetics of Ubx are very well known (Lewis, 1978; Kerridge We report a genetic and molecular study of UbxMX6 and and Morata, 1982; Casanova et al., 1985a) and there are Ubxl9Slxl, two mutations in the Ultrabithorax (Ubx) locus many viable mutations that are easy to characterize which appear to have a strong effect on the activity of phenotypically; (ii) there are many breakpoints known that the homologous Ubx gene. These mutations show the alter the pairing in the Ubx region; (iii) its molecular structure characteristic embryonic and adult phenotypes of Ubx is well known (Bender et al., 1983; Weinzierl et al., 1987; null , and also fail to produce any detectable Ubx O'Connor et al., 1988), which may open an avenue into the product. Yet, genetic and phenotypic analyses involving molecular analysis of transvection. a large number of trans heterozygous combinations of Ubx is a large gene, containing > 100 kb of DNA (Bender UbMxe6 and Ubx'95rxl with different classes of Ubx et al., 1983) including complex regulatory machinery. One mutations, indicate that they hyperactivate the transcription unit encodes all the functional proteins (Hogness homologous gene. This effect is induced on wildtype or et al., 1985) and at least two cis-acting regulatory regions, mutant forms of Ubx, provided that the pairing in the abx and bxd (Casanova et al., 1985a; Peifer and Bender, bithorax region is normal, i.e. these mutations have a 1986), are involved in the spatial expression of the strong positive effect on transvection. We also show that, transcription unit. These two regions act from a long distance unlike all the other known cases of transvection in Ubx, from the promoter of the transcription unit and have the this is not zeste-dependent. Southern analyses indicate properties of eukaryotic enhancers (Bender, 1985; Hogness that UbxmX6 is a 3.4 kb deletion, and Ubxl951xl is an - 11 et al., 1985; Simon et al., 1990; Muller and Schaffner, kb insertion of foreign DNA, both in the promoter region. 1991). They control the activity of the Ubx gene in specific We speculate that the region altered in the mutations may body regions (Casanova et al., 1985a; White and Wilcox, have a wildtype function to ensure cis-autonomy of the 1985). regulation of Ubx transcription. Lewis (1954) described transvection for some mutant Key words: homeosis/transvection/Ubx phenotypes which are affected by the degree of pairing between homologous ; the phenotype of trans heterozygotes between bx34e and Ubxl, two mutant alleles at the Ultrabithorax locus, depends in part of the pairing Introduction of the chromosomal region where the Ubx gene resides. Flies Homologous chromosomes are intimately paired in somatic of genotypes bx34elUbxl and R(bx34e)IUbxl (where R means cells of insects, as is most dramatically shown in the polytene a rearrangement preventing pairing in the Ubx region) exhibit chromosomes of Drosophila and other dipterans. The different phenotypes, the latter being stronger. We now know functional significance of this phenomenon is not clear, but that the partial rescue observed when pairing was normal its general occurrence may suggest a role in the control of could not have been due to any product originating from gene expression. Furthermore, some recent reports suggest the Ubxl , which is a null i.e. unable to that somatic pairing may also occur in mammals, as indicated produce functional protein (Weinzierl et al., 1987). It follows by the possibility of somatic recombination (Tartof and that the functional level of the Ubx gene in the bx34e Henikoff, 1991). chromosome depends on pairing, thus suggesting a functional Transvection is a phenomenon, originally defined by Ed cooperation between homologous genes. Lewis (1954) (reviewed in Judd, 1988; Wu et al., 1989; Transvection at Ubx is not restricted only to combinations Tartof and Henikoff, 1991), that suggests a functional role involving the bx34e allele, which is a mutation in the abx for somatic pairing. Although a large body of evidence regulatory element; it has also been described for other (abx indicates that pairing is not a critical requisite for normal and bx) mutant alleles in the same regulatory region and also gene function-since rearrangements do not detectably affect for mutants in the bxd regulatory element (Lewis, 1982, gene activities-transvection points to the possibility of 1985; Mathog, 1990). regulatory interactions between genes situated in homologous Other cases have been described that can also be explained chromosomes (Judd, 1988; Tartof and Henikoff, 1991), and in terms of trans interactions between homologous its study may contribute to elucidate the mechanism of action chromosomes. For example, the dominant mutation of cis-regulatory elements (Zachar et al., 1985; Geyer et al., Contrabithoraxl (Cbxl) produces a partial transformation of 1990). It has been described for white (Zachar et al., 1985), the wing into haltere (Lewis, 1963; Morata, 1975; Casanova decapentaplegic (Gelbart, 1982), brown (Dreesen et al., et al., 1985b), caused by ectopic expression of the Ubx Oxford University Press 3645 A.Martinez-Laborda, A.Gonzalez-Reyes and G.Morata

unit in the where it is not A) transcription wing cells, normally ubx b&Xd expressed. This abnormal expression is produced by a regulatory regulatory transposition of regulatory sequences within the Ubx gene region region (Bender et al., 1983; Casanova et al., 1985a; Karch et al., 1985). It can be shown that in Cbx', the regulation of the two homologous Ubx genes is altered; the suppression of the cis activity of Ubx in Cbx1Ubxll+ flies (Ubxl is null) does not completely eliminate Ubx ectopic expression in the wing cells (and hence the Cbx phenotype), indicating an ectopic activation of the wildtype homolog. Furthermore, the trans heterozygote CbxllUbxl shows less transformation than Cbxll + because of the elimination of the Ubx function B) of the trans chromosome. Thus alteration of the cis- regulatory region of the Cbxl chromosome has both a cis and a trans effect. In conclusion, transvection is a general phenomenon within the Ubx gene. Although there are differences depending on the mutation under test, all the regulatory regions of the gene can be shown to be affected by pairing. In this paper we describe two new mutants in the Ubx locus which have a strong positive effect on transvection. The two mutations (a deletion and an insertion) map around the promoter region Fig. 1. (A) Molecular map of the Ubx gene showing the mutations of the Ubx transcription unit. We propose that this region used in this work. Deletions are represented as rectangles and may act as a negative regulator of transvection. insertions as triangles. The main Ubx RNA is shown below, the exons appearing as solid boxes. Cbxl is an insertion of the DNA deleted in pbx'. Ubx1 is a Doc transposon inserted in the 5' untranslated region of Ubx. Ubx195 is a single base pair subtitution introducing a nonsense Results codon into the second microexon, and UbxMXJ7 is an inversion within the Ubx transcription unit including the second microexon. The dotted Isolation and molecular characterization of the triangles (bx34e, bx3 and bxd1) are gypsy insertions. (B) Molecular mutations Ubx'95rxI and UbxMX6 characterization of UbxJX6 and UbxIrr. Southern analysis showed abnormal restriction fragments when plasmids p3108 and p3102 were The UbxJ95rxJ allele was found after irradiation of a used. Restriction sites EcoRI (E), Hindlll (H), MluI (M), NruI (N) chromosome carrying the point mutation Ubx'95, in a and Sall (S) are indicated. The transcription start site is indicated by a strategy designed to isolate other mutations that rescue its black dot and the arrow shows the direction of transcription. In null phenotype. Mutagenized chromosomes were tested over UbxI95sxJ the lesion was first located in the I kb EcoRI-HindIII induced- fragment containing the transcription start site. A new Sall fragment is UbxM4, a weak ethylmethane sulphonate (EMS) detected with both p3102 and p3108, indicating that the mutation allele isolated by E.Sanchez-Herrero, that allows for adult consists of an insertion of - 11 kb without any Sall restriction site. viability. Heterozygous UbxJ95IUbxM4 flies display a slight More precise localization involved single and double digestions using but clear transformation of haltere (a T3 appendage of the MluI, NruI and PstI. The insert is located in a 317 bp fragment adult) towards wing (the corresponding appendage of T2). defined by MluI and NruI, which also contains the transcription start. In UbxAX6 the EcoRI and HindIII sites flanking the transcription start In addition, the first abdominal segment is partially site are absent, and a new EcoRI-Hindlll fragment of 2 kb appears transformed into thorax. A mutation called UbXJ95rXJ was with both p3102 and p3108. These results indicate a deletion of -3.4 isolated that shows virtually wildtype phenotype over kb. UbxM4. The other mutation was X-ray induced on a chromosome the AluI restriction pattern in the second microexon genomic already carrying the mutation abd-AMJ [an amorphic allele region (Weinzierl et al., 1987). of the abd-A gene (Morata et al., 1983; Sanchez-Herrero et al., 1985)] and named Ubxmx6 (Casanova et al., 1987). Genetic characterization of Ubx'95rxl and UbxMX6 This was described as an amorphic allele of Ubx in embryos, Neither ofthe two alleles possesses embryonicfunction. Both but showing moderate adult mutant phenotypes in alleles, UbxJ95rxJ and UbxMX6, are homozygous and combination with viable mutations of the Ubx gene hemizygous lethal. UhxJ95rxl mutant embryos show the (Casanova et al., 1987). characteristic syndrome described for Ubx null alleles The molecular characterization of the two alleles was (Lewis, 1978; Hayes et al., 1984); parasegments 5 and 6 carried out by Southern analysis and the results are illustrated are transformed into parasegment 4 and there is a slight in Figure 1. The Ubxmx6 mutation consists of a 3.4 kb thoracic transformation of the A2 - A7 abdominal segments. deletion that includes the HindIm -EcoRI fragment harboring In the case of the Ubxmx6 chromosome, which is also the two adjacent transcriptional start sites of Ubx (Saari and deficient for abd-A function, we observed the sum of Ubx Bienz, 1987) and adjacent DNA. and abd-A mutant phenotypes, just as in Df(3R)Ubx'09, a The Ubxl95rxl allele carries an -11 kb insertion in the deletion of both Ubx and abd-A genes (Morata et al., 1983; same 1 kb HindIlI-EcoRI fragment mentioned above. Casanova et al., 1987). The trans combination UbxMX6 Further subdivision allows the localization of the insert in abd-AMJ/Df(3R)bxdJ00, in which there is normal abd-A a smaller 317 bp fragment defined by the restriction sites function, showed null Ubx phenotype. of NruI and MluI (Saari and Bienz, 1987). In addition, this In agreement with the embryonic phenotypes, Ubxl95rxJ mutation contains the Ubx'95 point mutation as detected by and Ubxmx6 homozygous embryos lack detectable levels of 3646 Trans interactions in Ubx

Ubx protein as measured by a specific anti-Ubx antibody (White and Wilcox, 1984). Neither ofthe alleles has imnaginalfifnction. We have studied the adult phenotype of the two alleles by generating hemizygous or homozygous cell clones for each of the alleles in heterozygous animals. The clones were produced by means of X-ray induced mitotic recombination as detailed N;. in Materials and methods. Two examples are shown in Figure 2. Mutant clones for Ubxl95rxl and UbxMX6 were produced during the larval period, 72-96 h after egg laying. Many clones were found in all body regions. Those in the halteres, the three sets of legs and the abdominal segments were examined for homeotic transformations, as they derive from the body region where Ubx is expressed. For UbxJ95rvJ we found a total of 22 clones in the haltere segment. All of them showed a complete transformation towards the homologous wing structure; those in the haltere appendage were transformed into wing, and those in the metanotum showed a mesothoracic transformation (Figure 2a). Since these clones were labelled with mwh, which marks individual trichomes, we could in many cases determine that all the cells were homeotically transformed. This is characteristic of null Ubx mutants (Morata and Garcia-Bellido, 1976; Kerridge and Morata, 1982). Clones appearing in the third leg (12) exhibited a transformation into the homologous region of the second leg (Kerridge and Morata, 1982). In the posterior compartment of the second leg they differentiated normally, as expected because the ppx transformation appears only in early induced clones (Morata and Kerridge, 1981). UbxJ95rxl mutant clones did not materialize in the first abdominal segment, another typical Fig. 2. (a) Clone (arrow) in the metanotum of UbxMX6 homozygous feature of Ubx- clones, probably because they are cells marked with Sb', showing a mesothoracic transformation (see Materials and methods for details). (b) Large clone (arrow) of transformed into thorax, and thoracic development cannot UbxI95rxI hermizygous cells also showing mesothoracic transformation proceed in an abdominal environment (Morata and Garcia- in the metanotum. Bellido, 1976). In the rest of the abdominal segments,

UbxJ95rxl mutant clones differentiate normally. studied and because it carries a lesion consisting of a single The results obtained with UbxMX6 clones closely parallel base pair substitution (Weinzierl et al., 1987) producing a those of UbxJ95rrI; one example is shown in Figure 2b. The stop codon that eliminates >90% of the Ubx proteins. only difference is that mutant clones failed to appear in all Pairing is therefore not disrupted in combinations involving the abdominal segments. This is because these clones are Ubx'95 and so transvection should be normal. Furthermore, also mutant for abd-A, and this results in thoracic except for the ppx transformation, the phenotypes it produces development of all the abdominal segments. are like thoseUbxtof null mutations such as and UbX922 In summary, the results obtained with the clonal analysis (Kerridge and Morata, 1982). strongly suggest that both UbxJ95rxl nor UbxMX6 behave as Trans heterozygotes of Ubx195, Ubx'95r'l. and Ubxp6uwith null alleles for imaginal function. abx2, bx34e, bx3, pbxl and pbx2, as well as with weak Ubx alleles affecting the entire domain (Ubxiaon,Ubxm4 and Ubx19-5r1 and UbxMX6 partially rescue the phenotype Cbx'), were constructed and their phenotypes were of viable mutations in the Ubx locus analyzed and quantified. Some of these results are ilustrated Four individual homeotic transformations are associated with in Figure 3. These can be summarized by saying that the the Ubx mutant syndrome (Casanova et al., 1985a): homeotic transformations in the combinationsaiof UbxI9r postprothorax (ppx), bithorax (bx), postbithorax (pbx) and and Ubxmx6 are much weaker than those of Ubx195. The bithoraxoid (bxd), affecting the four compartments T2p, T3a, only exception is the ppx phenotype in which there is no T3p and Ala respectively, which constitute parasegments difference from Ubxc95; the ppx transformation in 5 and 6. The ppx and the bx transformations result from lack UbxMX6Iabx2 is 48%, similar to the figures reported for of function of the abx cis-regulatory unit, and bxd and pbx strong Ubx alleles or the deletion of the gene (Casanova are produced by a defect in the bxd cis-regulatory unit et al., 1985a). In the case of Ubxmmari the penetrance of the (Casanova et al., 1985a). Some of these transformations are ppx transformation is 5%, similar to that found for Ubx'95 shown in Figure 1. (Weinzierl et al., 1987; our own results). This low We have analyzed the phenotypes of UbxJ9Sr-r and penetrance inUbxf95 may indicate that this mutation retains UbxMX6 in trans with several mutant alleles. We used the some Ubx function, even though the phenotypes it produces t Ubx'95 allele as a reference for the adult transformations inxtransfcormbiationis other and d mr tation because it is the parental chromosome of one of the mutations are as strong as those produced by null alleles. We note that 3647 A.Martinez-Laborda, A.Gonzalez-Reyes and G.Morata

.;

... iX ,. g .S

,}, *t R

k

¢,i. d; r i!ZR . :>: v .:

:. I :. 1 4 i,

.. i, .1

..s^.fX 4EeS;

.{ E ......

i iS,\x

e;i+. w.

,- ;..

Fig. 3. Comparison of the phenotypes of some combinations of Ubx195 and UbXMX6. (a) bx.x3'/1Ub.xC195: the haltere appendage is partially transformed towards wing, as indicated by its larger size and the presence of bristles characteristic of the anterior wing margin. Note (arrow) the metanotum, a narrow piece of cuticle completely devoid of bristles. (b) R(bA34e)1Ubx195: the loss of transvection does not affect the transformation in the appendage; the difference is in the metanotum, which is partially transformed into mesonotum..,:t-as :indicated.. by the thoracic bristles (arrows). (c) bx24eIUbxMx6: the transformation in the haltere is virtually eliminated. (d) R(b.ri4e)1Ub.MxX6: the loss of transvection results in a transformation like that shown in panel b, including the thoracic bristles in the mesonotum (arrows). Comparison of panels c and d clearly shows the hyperactivity induced by UbxMX6 on the bx34' chromosome. (e) Ub.vX'7/UbxI95: in this genotype, the haltere appendage is transformed into wing in the anterior (bx transformation) and posterior (pbx) regions. In addition, the first abdominal segment (b.xd transformation) is missing. The second abdominal segment (II) is unaffected. (f) Ub_XMX17IUbXMX6: the haltere appendage is much more normal and the first abdominal segment (I) is restored. (g) pbxl/UbxMX6: the haltere appendage is mildly transformed towards wing in the posterior region. (h) R(pbx.l)Ub.vMX6: the loss ot' transvection results in a strong transformation of the posterior haltere region into posterior wing.

3648 A.Martinez-Laborda, A.Gonzalez-Reyes and G.Morata the rescue is always greater in combinations containing flies is like that of z+;bx34eIUbxMx6, essentially wildtype, Ubxl95rrlI, which is able to rescue completely a strong allele although sometimes the halteres of flies of the z- like bx3. combination are slightly bigger than in z+. The same result All these observations indicate that the UbxI95rdI and was obtained for the combinations of the UbXJ95rxJ UbxMX6 mutations induce extra activity in the homologous mutation. This is the first case of a transvecting combination Ubx gene. This excess of activity can also be noticed in of Ubx that is not z-dependent, and suggests that the comparison with combinations in which there is little or no mechanism of trans interaction with which we are dealing transvection; the combinations of Ubx'95 with pbx', bxdl, here differs from that of normal transvection. As a control, UbXMX17, UbxM4 and Cbx', in which pairing is normal, are we checked that transvection in the Ubx195 allele is z- as strong in the haltere transformation as the non-transvecting dependent; the combination z- ;bx34eIUbxJ95 shows a ones of the same viable alleles with Df(3R)P9 or Ubx'30. phenotype stronger than that of z+ ;bx34eIUbx'95 and similar Yet, the same alleles produce much weaker or wildtype to that of R(bx34e)1UbxJ95. phenotypes in trans to Ubxl95rrll and UbxMX6 (see Figure 3e, f, g and h). Discussion UbxMX6and Ubx'95rxl hyperactivate a wildtype Ubx Transvection in UbxMX6 and Ubx'95rxl gene in trans We have characterized two lethal Ubx alleles with an unusual A general feature of the Ubx gene is that it is haplo- property. They are null mutations, unable to produce any insufficient, so that one dose of the gene is not able to functional product; yet, when they are normally paired with produce a normal haltere; a Ubx-l+ fly has enlarged a Ubx gene that is able to function, give rise to an extra halteres that contain one to five bristles, indicating a very activity of the trans gene. No other known Ubx allele behaves mild transformation towards wing. We note (Figure 4) that in this way. Southern analysis indicates that in both cases trans heterozygotes of the two transvecting alleles show a region of the Ubx promoter is affected. In one case a 3.4 virtually normal halteres, indicating an increased activity of kb sequence is deleted, and in the other an - 11 kb fragment the wildtype Ubx gene in trans. is inserted in the same region (Figure 1). Elimination or alteration of these sequences of Ubx impedes transcription, The extra activity of the Ubx gene in trans to but at the same time augments the activity of the homolog; Ubx'95rxl and UbxMX6 is pairing- but not zeste- transvection is increased. We have noticed this in all the dependent combinations we have analyzed; the Ubx function in the We have tested to what extent the extra activity seen in abx2, bx3, bx34e, pbx', Cbx', UbXMXI7 and UbxM4 chromosomes in trans to the new Ubx alleles depends on chromosomes is increased with respect to that of the same pairing. To this end, several rearranged (R) chromosomes chromosomes in trans with regular null Ubx alleles. carrying mutations and breakpoints preventing pairing in the Moreover, we observe that the typical haplo-insufficient Ubx region were used. The basic observation is that the lack phenotype of null Ubx alleles (Ubx-l+ flies show an of pairing results in the loss of the rescue activity of the two engrossment of the haltere) is not present in Ubxmx6/ + and new Ubx alleles, and is illustrated by the comparison of Ubxl95rxII + flies, indicating that not only mutant Ubx Figure 3c and d, and of Figure 3g and h. While the genes, but also the wildtype shows increased activity. phenotype of UbxMx6Ibx34e is virtually wildtype, that of Of the combinations that we have studied, those involving UbxMX61R(bx34e), in which a translocation prevents normal bx34e and pbxl are the best documented because we can pairing in the Ubx locus, exhibits a strong homeotic study their function in trans with the different Ubx mutations transformation. The dramatic difference between these two in pairing and non-pairing conditions. These two mutations phenotypes emphasizes that the mutant UbxMX6 gene are caused by defects in cis-regulatory regions located far induces a great deal of activity in the homologous gene and apart; bx34e is a gypsy insert 60 kb from the 17 kb deletion that this activity is entirely dependent on appropriate pairing. causing the pbxl phenotype. These alterations are located A similar observation can be made for the set of in regulatory sequences that have all the properties of combinations of the two Ubx mutations with pbx' and transcriptional enhancers (Simon et al., 1990; Muller and R(pbxl) (Figure 3) and with pbx2 and R(pbx2). Bienz, 1991), and the fact that both Ubxmx6 and UbxJ95rxl We also note that the suppression of the haplo-insufficient suppress their mutant phenotype strongly indicate that their phenotype of UbxMX6 and UbxI95r-,l is pairing-dependent. enhancers are substituting for the defective ones of the bx34e For example, elimination of the pairing of Ubxl95rrl with and pbx' chromosomes. Thus in the two Ubx alleles the the wildtype homolog in flies of genotype DpP]151+; alterations of the promoter region appear to have a long range Df(3R)PJJS/Ubx'95r-'l, restores the typical haplo-insufficient effect. It is consistent with the current thinking (Geyer et al., phenotype of Ubx null alleles (Figure 4). In these flies, the 1990; Muller and Schaffner, 1990) about transvection, that wildtype allele of Ubx is in Dp P115 in the first chromosome invokes the effect of enhancers of one gene on the and unable to pair with the Ubx195rrJ gene. transcription of the homolog. Since in other combinations of the BX-C the partial One aspect worth noting is that neither of the two new complementation due to transvection depends on the normal alleles suppresses the early ppx phenotype (Casanova et al., function of the zeste (z) product (Kaufman et al., 1973; 1985a). This phenotype has been shown to result (Struhl, Micol and Garcia-Bellido, 1988), we tested whether the extra 1982) from an inappropriate activation of the Scr gene in activity induced by our Ubx alleles also requires it. We used the T2p compartment due to the loss of a repressing activity the null alleles Z_69-2 and z69-3 (Gelbart and Wu, 1982). To of Ubx at the embryonic period, before 10 h of development. our surprise, the lack of z function does not affect the Indeed, in abx2 mutants the second leg imaginal discs phenotype; the transformation observed in z-;bx34elUbxM'6 contain Scr protein which is not present in the wildtype

3649 A.Martinez-Laborda, A.Gonzilez-Reyes and G.Morata

A

abx Wlldtype _- rtransvectiont not detectable

bx 34e abx Weak transvection 195_zeste-dependentenhancer moter Ubx3953.

B

5, promoter 3 "7Gipsy bx 34e ij - strong transvection abx enhancer rI) zeste-independent 11- IIAV _i UbxYVLIA.-x6-T- 1/-- 3 twj=wam

-.----r>-promoter 3. + abx 7I11 strong (?) transvection enhancer zeste-independent - _ uuxI)hbxl1x6A 3, W19=19M

Fig. 5. Interpretation of some of the transvecting combinations Fig. 4. Hyperactivation of the wildtype Ubx gene by UbxJ95r.CI and its involving bx4e. In the wildtype (A, top), enhancer-like elements such dependence on pairing. (a) UbxI95rxJI+ haltere, showing suppression as abx would act on their cis promoters. Any possible trans interaction of the typical haploinsufficient phenotype of null Ubx alleles. (b) would probably go unnoticed. In bxi4elUbxlY (A, bottom), the abx DpPI151+;UbxI9srxIlIDf(3R)PJ15 haltere: the Ubx haplo-insufficient enhancer of the Ubx195 chromosome is able to transactivate weakly the phenotype reappears due to the loss of pairing in the BX-C region. Ubx transcription unit of the bX34, chromosome, partially rescuing the mutant phenotype (compare Figure 3a and b). In bxi4elUbx2IX6 (B, top) the same enhancer is able to act much more effectively on the bx34e (Littley et al., 1990). As the ppx function is associated with chromosome, producing a near wildtype phenotype (Figure 3c). The the abx regulatory element which is also responsible for the same abx enhancer of the UbxMX6 gene can also act (B, bottom) on a bx transformation, suppressed by UbxI95rxJ and UbxMX6, we wildtype Ubx gene producing an excess of activity that suppresses the haplo-insufficient phenotype (see Figure 4a and b). suppose that these are unable to suppress ppx because the hyperactivation of the homolog occurs after 10 h of development. A mechanism for preventing transvection? The combinations of bx34e are of special interest, for they Transvection was originally described for trans interactions indicate that there are different levels of transvection. in Ubx (Lewis, 1954) and involves a gene on one Compare the phenotypes of UbXJ95Ibx34e and chromosome being able to influence expression of its paired UbxI95IR(bx34e) (Figure 3a and b), a typical case of homolog; this effect disappears if pairing is eliminated. Other transvection. There is a partial rescue of the Ubx'95/R(bx34e) Drosophila genes, including w, br, dpp and y, also show phenotype in the Ubx'95Ibx34e flies due to higher activity of transvection. What is the significance of transvection? For the bx34e chromosome. However, in the combinations of most genes, including those that transvect, the lack of pairing bx34e with UbxMX6 (Figure 3c) and UbxI95rxI, the phenotype does not have a detectable effect. A functional copy of Ubx, is virtually wildtype, despite the fact that the two mutations as in the case of Ubx1301+ flies (non-transvecting, pairing are null, indicating a much higher activity of the Ubx gene at Ubx completely eliminated), is all that is needed to obscure carrying the bx34e mutation. Thus the two Ubx alleles cause any subtle difference with UbxlI + ones (transvecting, a level of transvection which is much stronger than the pairing normal). By this sort of test, transvection appears normal one (Figure 5). Moreover, this elevated level of to be of little significance. However, the detailed study of transvection does not depend on the presence of the zeste some cases of transvection may shed light on the regulatory product, suggesting that the mechanism altered in the two mechanisms operating in large genes. The best documented new mutants differs from that operating in normal cases of transvection involve interactions between mutant transvection. forms of a gene lacking some regulatory components. This The genetic properties of UbxMx6and UbxI95rxl suggest a is so not only for Ubx, but also for yellow. The latter is quick genetic method to detect lesions in the promoter region particularly interesting as there is a detailed molecular of preexisting Ubx null mutations, as they will rescue totally characterization of the transvecting combinations. Geyer or partially the phenotype of strong combinations such as et al. (1990) have demonstrated that a specific enhancer bx3/Ubx-. element of the yellow gene that cannot act in cis (because 3650 Trans interactions in Ubx a critical part of the promoter is deleted), can nonetheless phenomena are rare, even in Drosophila. Transvection is activate the trans homolog, defective in the enhancer. This normally detected in special mutant combinations, that results in virtual complementation, yielding a wildtype fly. because of the particular gene architecture they generate, These authors postulate that there are two requisites for the allow trans regulation to occur. This situation is clearly phenomenon: one gene must be defective for the promoter illustrated in the yellow gene, which usually does not and the homolog for the enhancer element. In normal transvect, but that under certain conditions can show strong circumstances, each enhancer will only act on its own transvection (Geyer et al., 1990). One might speculate that promoter, and as long as one of the homologs is intact, a mechanism to prevent trans interactions may be yellow function will not be affected by lack of pairing. Only biologically significant, and not only for dipterans, for when the enhancer of one gene and the promoter of the other eukaryotic chromosomes contain many DNA loops which are defective, can the trans phenomenon take place. This may be so close that enhancers of one gene might act on requires physical proximity and hence it is pairing dependent. the promoter of another. This would undesirably alter normal Our results with the new transvecting Ubx alleles fit well gene regulation and evolution may have developed a with the model of Geyer et al. (1990). Our view about mechanism to prevent it. transvection in Ubx is schematized in Figure 5, for bx34e, although it is probably applicable to the rest of the transvecting combinations of Ubx. We think that a critical Materials and methods element of transvection in Ubx is a region located at or near Fly stocks and culture conditions the promoter, which is defective in our two mutations. When Mutant alleles used in this work are referred to in the main text when present, the interaction between homologous chromosomes appropriate. Most of them have been previously described (Lindsley and weak (as in flies, Figures 3 and SA) or non- Grell, 1968; Lewis, 1978; Sdnchez-Herrero et al., 1985; Peifer and Bender, is bxI4elUbx'95 1986; Casanova et al., 1987; Weinzierl et al., 1987; Micol and Garcia- existent, but in its absence, any properly paired Ubx Bellido, 1988; Busturia et al., 1990). A description of the deletions Df(3R) homolog, whether mutant or wildtype, will become P9, Df(3R) bxdl°° and Df(3R) Ubx109 can be found in Lewis (1978) and hyperactive (as in bxI4elUbxMX6 flies, Figures 3 and 5B). Morata et al. (1983). Rearranged chromosomes used to test the transvection This is supported by the phenotype of pbxl combinations effect are the following: Tp(3,3) P47 [Tp(3,3) 66B; 89D; 92A], R(bx34e) in the text; In(3LR) P88 [Inv(3LR) 61A1-2; 89C2-4], R2(bx4e); T(2,3) [which are not rescued by transvecting Ubx alleles like bw'De3 [T(2,3) 59D; 81FJ, R(pbxl); they have been described previously Ubxl or Ubx195 in which the region around the promoter is (Lindsley and Grell, 1968; Castelli-Gair, 1989). intact] but are partially rescued by Ubxmx6 and UbxJ95rxJ. Flies were cultured on standard media and under uncrowded conditions The strong positive effect of UbxJ95rxJ and Ubxmx6 on at 25°C, except crosses with abx2 which were done at 17°C to increase transvection can be explained by a phenomenon of promoter the penetrance of the ppx transformation (Casanova et al., 1985a). competition, as in the case of yellow (Geyer et al., 1990). Phenotypic analysis Normally, distant cis-regulatory elements of Ubx would Many of the results presented in this paper involve comparing the phenotypes interact with their own promoters, which are physically of different genetic combinations in the presence or absence of pairing. In closer or are more accessible. When these enhancers are not all the cases considered the difference is qualitatively obvious, although in many of them we have quantified the transformations in order to facilitate able to act in cis because the promoter is missing or is the comparison. Nevertheless, we judge that it is not necessary to present functionally inactive, the action in trans would be permitted, all the numerical values found for each combination. Most of the phenotypes giving rise to a trans interaction (transvection). The problem involve the transformation of halteres, which have no bristles, into wing, with this kind of explanation is that it is hard to imagine which contains long rows of bristles in the anterior (87 6) and in the posterior (212 4 21) margins. Thus the transformation towards wing can how a long range acting enhancer, abx, which is - 30 kb be measured as the percent of bristles in the mutant haltere with respect away from the Ubx promoter, can distinguish its own to the number in the wing. A similar measure can be given of the promoter from that of the homologous gene, especially in transformation of metanotum, with no bristles, into mesonotum, which view of the intimate pairing of homologous genes in contains an average of 114 + 9 bristles. Drosophila; the observation that there is intracistronic mitotic Clonal analysis recombination (Lawrence and Green, 1979) indicates that Mutant clones for Ubx195r, were generated by X-irradiation (1000 rad) two homologous genes must be physically very close. of second and third instar larvae of genotype y;Dp(1;3)scJ4Dp(3;3)146 Consider, for example, the lack of complementation in M(3)i55 Df(3R) P1151mwh jv Ubx1 Srr.J In these larvae, a mitotic genotypes like UbxJ951bx34e, in which the normal abx recombination event in the left arm of the third chromosome proximal to normal promoter of the Dp (3;3) 146, which carries a normal dose of the BX-C, results in clones enhancer of Ubx'95 cannot act on the of cells marked with y, mwh, jv and M+ which are hemizygous for the bx34e, for if it did, it would have resulted in a wildtype Ubx mutation; as these cells lose the retarding Minute condition, they phenotype. Thus the abx enhancer of Ubx'95 recognizes and proliferate faster than the surrounding cells [see Morata et al. (1983) for selects its own cis promoter, located 30 kb away. details of this method of mitotic recombination]. From our results one could also speculate that the For UbxMX6, we constructed and irradiated larvae of genotype UbxMX6 abd-AMIlKi Sb63 M(3) w124 to generate Ki+ Sb+ UbxMXOabd-AMIM+ cell sequences missing or inactivated in our mutations have a clones. The fact that the clones were also defective in abd-A function is positive role in preventing transvection, that is, in impeding irrelevant for their differentiation in the metathorax and first abdominal inappropriate control of one Ubx gene by the regulatory segment because abd-A is not expressed in those segments. Since Sb63 and machine of its homolog. Since the somatic chromosomes of Ki are not reliable markers for the first abdominal segment, we also X- irradiated third instar larvae of genotye y;Dp(J;3)sc'4 Dp(3;3)146 M(3)i55 dipterans are intimately paired and much of gene regulation Df(3R)PJJS/mwh jv UbxX6abd-A to produce clones marked with y, involves enhancers acting at long distances, in some instances mwh, jv and M+ (Morata et al., 1983) which are readily scorable in the the regulatory regions of one gene can act on the homolog, first abdominal segment. as in the transvection phenomenon. Therefore a mechanism might have developed to ensure cis-autonomy, so that each Preparation of the larval and adult cuticle If For the larval cuticle we used the method of Van der Meer (1977), slightly gene is controlled by its own cis-regulatory sequences. modified by using Hoyer mountant diluted 1:1 with lactic acid (Wieschaus this 'antitransvection' mechanism were to exist, it would and Nusslein-Volhard, 1986). Adult cuticle was prepared by cutting the explain why, despite intimate pairing, transvection appropriate pieces under the dissecting microscope. The internal organs were 3651 A.Martinez-Laborda, A.Gonz6lez-Reyes and G.Morata

digested with hot 10% KOH, and the cuticle was washed in alcohol and Lindsley,D.L. and Grell,E.H. (1968) Genetic Variations of Drosophila mounted in Euparal. melanogaster. Carnegie Institute, Washington, publication 627. Littley,J., Byrd,C. and Brower,D. (1990) Genetics, 124, 899-908. Antibody staining Macias,A., Casanova,J. and Morata,G. (1990) Development, 110, We used the monoclonal anti-Ubx antibody developed by White and Wilcox 1197-1207. (White and Wilcox, 1984). Antibody staining was done using the Vectastain Mathog,D. (1990) Genetics, 125, 371-382. ABC kit as previously reported (Macfas et al., 1990). Embryos were Micol,J. and Garcia-Bellido,A. (1988) Proc. Natl. Acad. Sci. USA, 85, dehydrated in an alcohol series and mounted in Araldite. 1146-1150. Micol,J., Castelli-Gair,J. and Garcia-Bellido,A. (1990) Genetics, 126, Southern analysis 365-373. For DNA extraction, flies were homogenized in hydration buffer (0.1 M Morata,G. (1975) J. Embryol. Exp. Morphol., 34, 19-31. NaCl, 0.2 M sucrose, 10 mM EDTA, 30 mM Tris pH 8.0, 0.5% Triton Morata,G. and Garcia-Bellido,A. (1976) Wilhem Roux's Archives, 179, X-100). The homogenates were filtered and centrifuged and the pellet was 125-143. resuspended in lysis buffer (10 mM Tris pH 8.0, 10 mM EDTA, 0.35 NaCl, Morata,G. and Kerridge,S. (1981) Nature, 290, 778-781. 1 % sarcosyl). After adding proteinase K, samples were incubated for 2 h Morata,G., Botas,J., Kerridge,S. and Struhl,G. (1983) J. Embryol. Exp. at 37°C and extracted three times with phenol:chloroform (1: 1). The nucleic Morphol., 78, 319-341. acids were precipitated twice with ethanol and resuspended in TE (10 mM Mfiller,J.B. and Bienz,M. (1991) EMBO J., 10, 3147-3155. Tris pH 8.0, 1 mM EDTA) with 20 Agg/ml boiled RNase A. Muller,H.P. and Schaffner,W. (1990) Trends Genet., 6, 300-304. Southern blots and synthesis of labelled DNA probes were made as O'Connor,M., Binari,R., Perkins,L. and Bender,W. (1988) EMBO J., 7, described in Sambrook et al. (1989). Exposures were at -70°C with 435-445. intensifying screens for 3 days. Peifer,M. and Bender,W. (1986) EMBO J., 5, 2293-2303. Saari,G. and Bienz,M. (1987) EMBO J., 6, 1775-1779. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning. A Acknowledaements Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring We thank Isabel Vern6s and Jordi Casanova for their contribution in the Harbor, NY. early part of this work, Jordi Casanova, Jaime Castelli-Gair and Sanchez-Herrero,E., Vernos,I., Marco,R. and Morata,G. (1985) Nature, Ernesto 108-113. Sanchez-Herrero for their helpful comments on the manuscript, and Jaime 313, Castelli-Gair for communicating results prior to We also Simon,J., Peifer,M., Bender,W. and O'Connor,M. (1990) EMBO J., 9, publication. thank 3945 Sigrid Baars for her help with the Southern analysis. This work was supported -3956. by grants from the DGICYT and the Fundaci6n Ram6n Areces. Struhl,G. (1982) Proc. Natl. Acad. Sci. USA, 79, 7380-7384. Tartof,K. and Henikoff,S. (1991) Cell, 65, 201-203. Van der Meer,J. (1977) Dros. Inf: Serv., 52, 160. References Weinzierl,R., Axton,J., Ghysen,A. and Akam,M. (1987) Genes Dev., 1, 386-397. Babu,P. and Bhat,S. (1981) Mol. Gen. Genet., 183, 400-402. White,R. and Wilcox,M. (1984) Cell, 39, 163-171. Babu,P., Selvakumar,K.S. and Bhosekar,S. (1987) Mol. Gen. Genet., 210, White,R. and Wilcox,M. (1985) Nature, 318, 563-566. 557-563. Wieschaus,E. and Nusslein-Volhard,C. (1986) Drosophila: A Practical Bender,W. (1985) Cell, 43, 559-560. Approach. IRL Press, Oxford. Bender,W., Akam,M., Karch,F., Beachy,P., Peifer,M., Spierer,P., Wu,C., Jones,R., Lasko,P. and Gelbart,W. (1989) Mol. Gen. Genet., 218, Lewis,E. and Hogness,D. (1983) Science, 221, 23-29. 559-564. Busturia,A., Vernos,I., Macias,A., Casanova,J. and Morata,G. (1990) Zachar,Z., Chapman,C.H. and Bingham,P.M. (1985) Cold Spring Harbor EMBO J., 9, 3551-3555. Symp. Quant. Biol., 50, 337 -346. Casanova,J., Sanchez-Herrero,E. and Morata,G. (1985a) Cell, 42, 663-669. Received on April 15, 1992; revised on June 25, 1992 Casanova,J., Sanchez-Herrero,E. and Morata,G. (1985b) J. Embryol. Exp. Morphol., 90, 179-196. Casanova,J., Sanchez-Herrero,E., Busturia,A. and Morata,G. (1987) EMBO J., 6, 3103-3109. Castelli-Gair,J. (1989) Regulaci6n cis y trans de la expresion del gen Ultrabithorax de Drosophila melanogaster. Thesis. Universidad Aut6noma de Madrid. Castelli-Gair,J., Micol,J. and Garcia-Bellido,A. (1990) Genetics, 126, 177-184. Dreesen,T.D., Henikoff,S. and Loughney,K. (1991) Genes Dev., 5, 331-340. Gelbart,W. (1982) Proc. Natl. Acad. Sci. USA, 79, 2636-2640. Gelbart,W. and Wu,C.F. (1982) Genetics, 102, 179-189. Geyer,P., Green,M. and Corces,V. (1990) EMBO J., 9, 2247-2256. Hayes,P., Sato,T. and Denell,R. (1984) Proc. Natl.Acad. Sci. USA, 81, 545-549. Hogness,D., Lipshitz,H., Beachy,P., Peattie,D.A., Saint,R.B., Goldschmidt-Clermont,M., Harte,P.J., Gavis,E.R. and Helfand,S.L. (1985) Cold Spring Harbor Symp. Quant. Biol., 50, 181-194.

Judd,B. (1988) Cell, 53 , 841-843. Karch,F., Weiffenbach,B., Peifer,M., Bender,W., Duncan,I., Celniker,S., Crosby,M. and Lewis,E. (1985) Cell, 43, 81-96. Kaufman,T.C., Tasaka,S.E. and Suzuki,D.T. (1973) Genetics, 75, 299-321. Kerridge,S. and Morata,G. (1982) J. Embryol. Exp. Morphol., 68, 211-234. Lawrence,P.A. and Green,S.M. (1979). Dev. Biol., 71, 142-152. Lewis,E. (1954) Am. Nat., 88, 225-239. Lewis,E. (1955) Am. Nat., 89, 73-89. Lewis,E. (1963) Am. Zool., 3, 33-56. Lewis,E. (1978) Nature, 276, 565-570. Lewis,E. (1982) Control of Body Segment Differentiation in Drosophila by the Bithorax Gene Complex. Liss, New York. Lewis,E. (1985) Cold Spring Harbor Symp. Quant. Biol., 50, 155 -164. 3652